Solving and Winning a Doomer Future
Building early infrastructure in preparation for a new world, post exponential AI.
The Premise
We live in a drastically changing world. Current state of affairs revolve around the market for superchips soaring, $12.2B+ being deployed into AI-related startups quarterly, and autonomous agents replacing over 8000 jobs monthly (Snowflake). Everywhere you look, processes are being automated, costs are being reduced, and industries are changing, all because of the rapidly growing sector we call Artificial Intelligence. PwC estimates that AI will boost the GDP of China by a little over 26% by 2030 and of North America by 14.5%. These estimates are based on current growth trends and yet - the world is nowhere near the pinnacle of the AI Revolution. Despite the countless opportunities present in the market today, ideas like quantum computing are right around the corner. Once we solve the core bottlenecks relating to compute and improve latency/price 10x, things we never thought imaginable are suddenly a reality. Industry expansion will decrease the number of unicorns, but increase the number of decacorns present. Infrastructure needs justify the number of incumbents to spiral upwards.
We’ve already begun to see the initial macro changes. With the number of AI assistants (~8.4B) to surpass the number of people in the world by the end of year, niche sectors aren’t the only verticals experiencing a shift. The macro economy is at a precipice. Whether it be Harvey coming for the legal ecosystem, Cognition creating agents to replace software engineers, or OpenAI killing hundreds of startups every ship, model improvement and agent learning are on an exponential trajectory. In contrast to the dot-com burst of the early 2000’s, the launch of ChatGPT was the Netscape kickstarter. However, even as massive CAPAX accelerates the boom, the benefit of open-source development takes our current revolution a step further. What’s happening now is marked by more durable business models, a faster speed of innovation, and an urge to replace cognitive functions. With improvement speeds speeding up and AI systems outperforming industry specialists, we’re heading into a new era.
Software will become commoditized and costs will go to $0.
Very soon, no one will have the need to create software as there will be no economic benefit. Companies currently run by 100,000 will be run by 100 people, making the agentic calls and ensuring the orchestration layers work. Once current organizations are well operated, people will have two choices: a) build new stuff or b) spend their time in other ways. Our brains are always trying to figure out the easiest way to do things, creating novelty is rarely the answer and thus, will choose not to build from the ground up. There will always be a shortage of people seeking to break the industry and create something new. We’re not just talking about Silicon Valley hacker homes. There are ~168 million people employed in the United States, as of April 2024, and certainly far fewer opportunities to implement practical software in the world. Likewise, if everyone can create software, there becomes an abundance in supply. Basic economics state that having far greater supply than demand drives utility down and forces the cost to $0. If people aren’t financially motivated to produce software, the majority won’t. Non-novel behavior is rarely rewarded in the race for progress. Although everyone will be able to create software, there will be less SWE’s in the future. Instead, revenue and value will be achieved in creative, craft based ways, towards which the masses will flock.
The Scenario
Software application pairs hand in hand with advancements in hardware. The question of generalist humanoid application isn’t if, it’s when. Fifteen years from now, humanoids will be able to output specialist human labor cheaper, quicker, and more efficiently. Traditional human tasks will lack need, too, as we’re seeing with the autonomous vehicle industry now.
In a way, it’s possible to draw parallels to the Industrial Revolution, when machines began replacing human labor in factory settings. To adapt, humans moved to cities and urbanized communities to further technological progress with their ability to learn and adapt to circumstances. This is no longer possible, as humans are no longer the best at implementing solutions. The better models get at intellectual activities, the more they can build better models to be better at intellectual capabilities in a recursive function. As the gap between top industry experts and everyone else gets wider, not only does the bar for entry increase, but AI agents become much better, cheaper, and faster than the general population at driving progress forward. Thus, there is a ceiling for the number of new opportunities in the ecosystem.
When the job shortage of all time occurs, the meaning of money and purpose fundamentally changes. The human race needs a new mission.
Current employment rates likely drop 10x to favor experts managing AI agents. In practice, most tedious office jobs cease to exist. People are forced to move into qualitative spaces or explore activities beyond work. On another front, the internet at its core won’t exist due to the volume of artificial content released. Centralized sources of information and local repositories will be far more valuable. We are moving towards a point where media hubs holding large breadths of data will cease to exist and hold value. By 2026, 90% of online content will be artificially generated. By the way, this is on track to happen regardless. If we as a society don’t transition off these broad medias, continue to find ways to progress as humans, and become lazy creatures of entertainment instead, we will soon die as a species.
The Solutions
At this critical point, better education and more incentives to work don’t help. Instead, separating labor from earnings and establishing a craft-based economy may be the solution. Creating a lack of scarcity is first priority as wage and skill gaps widen.
Policy changes will have difficulty navigating shifts involving negative income tax and constructing a basic living wage. Some jarring changes will be enacted and built by someone. Therefore, these next few years will be the primary opportunities to build the ‘picks and shovels’ of the future. Think of this time as the opportunity to build the infrastructure for Uber before the cell phone was invented.
Any innovative discovery needs an equal input to quality output. As such, progress acts in parallel to resource maximization and discovery. Advancing the future of human behavior requires making the most out of current resources and exploiting others in the ecosystem. For example, the obvious answer is asteroid mining, made popular by ex SpaceX and NASA employees seeing an entry point into the market (they obviously haven’t seen the movie Don’t Look Up). Although current incumbents are stretched too thin to prioritize this effort, they still control vehicles traveling in and out of orbit, thus holding a monopoly on efforts to mine. There is an opportunity to create hardware that expedites mining practices, yet the giants would control utility. Perhaps well-backed contrarian companies like Varda stand a chance due to their differential approach of orbital reentry. Another outlier solution involves maximizing Earth’s resources through building cities on water. Following trends related to rising sea levels, this allows for organic material production, oil extraction (think of expanding oil rigs), and inhabiting a greater percentage of Earth. Strong tailwinds and institutional backing supports repeat founders like Elon Musk or Jeff Bezos to take on the task.
Once resource allocation is sorted, norms will shift towards socially accepted jobs in industry that will need to be anchored by changes in policy. Niche specialists will control portions of the market. With distributional advantages, smart generalists in numerous industries will hold the most power.
Solving bottlenecks in diverse, unrelated sectors sets the premise for macro change.
A quick glance into some niche solutions… and what winners will look like:
Financial: The adoption of mass cryptocurrency is needed from both a consumer and agentic front; to process transactions and maintaining record.
As the gap between the elite and the rest of the world grows, governmental payment rails become monopolies. Accessible payment rails reduce cost and prevent the misalignment of power. People need full control of their assets in a craft-based economy. The winner will capture 100M+ people on a gamified, high-retention platform and lock in larger enterprises incentivized by reach and exposure, aligning values. Being the company to take individuals to their first cryptocurrency transaction buys cognitive dissonance moving forward.
Consumer Practice: Losing physical computers is the next step in data visualization.
The abundance of online content means traditional internet consumption and entertainment loses value. The intersection between neurotech and AR/VR is the solution to changing the way we think about workflows. Current products like New.Supercomputer and Apple’s Vision Pro have a ceiling of improving productivity by 10-20%. With hypotheticals and pushbacks surrounding neural implants, creating lightweight hardware that can store large chunks of local data wins the industry. An innately difficult problem, hardware patents, and an unconventional launch builds moat over late-moving incumbents.
Health: The wearable market will remain saturated, yet the data provider(s) and underlying insurers may very well come from a unified source.
Targeting the root causes of aging and disease will revolutionize the world of healthcare. Rapidly falling birth rates prompt a clear issue: we, as a species, are currently at greater risk of extinction than overpopulation. Research focused longevity companies envision gold as big pharma incumbents struggle to diverge from drug discovery processes. For individuals, RL models will soon be good enough to compile personalized data relating to heart rate, blood sugar, and everything in between. Providing accurate assessment in real time supports preventative illness and broadens the market outside current users. Startups are at an advantage to aggregate cross-device data and build a data moat from the ground up.
Going down the funnel, change begets change. It’s clear that we are only at the top of the waterfall with AI integration continuously increasing in velocity. Recent press surrounding OpenAI’s safety team resigning is fantastic news for the continuous progress of foundational models. Bottlenecks surrounding constrained resources are now gone, allowing the team to ship at lightning speeds. However, due to the drastic shift this will create in all industries, distribution of power will increase in concentration. Short term, constant revenue and workforce integration is the primary motivator. Moving forward, the importance of aligned values in closed source models is of outmost importance as international competition increases (by the way, closed source beats open source). With everything scrutinized, there really isn’t that much risk to societal-ending AGI. Correctly incentivized companies operating the majority of capital are the solution to compounding long-term progress.
Loved this-its interesting to see how we’ll see agents and crypto interact in the future